LabVIEW

cancel
Showing results for 
Search instead for 
Did you mean: 

What is anti-aliasing?

I set up an FFT Power Spectrum and someone told me I may need to do anti-aliasing for some data I'm reading. I understand it has something to do with harmonics. Is this something that can be done in LabVIEW or is this a hardware issue?

Thanks =0)

Lauren
0 Kudos
Message 1 of 6
(3,308 Views)
Hi,
anti-aliasing is smoothing. This term is usually used in computer graphics. In graphics anti-aliasing is used to smooth some distortions on the pictures (for example big squares in zooming).
As every smoothing anti-aliasing is some kind of lowpass filter. So I think that advise was to perform some filtering to remove highfrequensy components from your signal.

Good luck.

Oleg Chutko.
0 Kudos
Message 2 of 6
(3,306 Views)
If your signal has frequency components that are higher than the Nyquest frequency (half your sampling rate), they will show up as alias frequencies that are lower. This can cause false results.

Your analog signal must be filtered such that only frequencyies suitable for your sampling rate are digitized.

Look at some of the information here at the NI site, e.g.:
http://zone.ni.com/devzone/conceptd.nsf/webmain/2EAC3B85027D8EE5862568C0005554AE?opendocument
or
http://sine.ni.com/apps/we/niepd_web_display.DISPLAY_EPD4?p_guid=B45EACE3DBE356A4E034080020E74861&p_node=DZ52008&p_submitted=N&p_rank=&p_answer=&p_source=External
0 Kudos
Message 3 of 6
(3,306 Views)
This article might help:
http://digital.ni.com/public.nsf/3efedde4322fef19862567740067f3cc/68f14e8e26
b3d101862569350069e0b9?OpenDocument

"uscmagz" wrote in message
news:5065000000080000000AD60000-1079395200000@exchange.ni.com...
> I set up an FFT Power Spectrum and someone told me I may need to do
> anti-aliasing for some data I'm reading. I understand it has something
> to do with harmonics. Is this something that can be done in LabVIEW or
> is this a hardware issue?
>
> Thanks =0)
>
> Lauren
0 Kudos
Message 4 of 6
(3,306 Views)
Ok, so concerning anti-aliasing and the Nyquist frequency. I am creating images using a rastering laser beam, basically I end up with a intensity graph that I change over to a jpeg image... there are periodic functions within the image that are measured that are periodic about 1 per 100 micron or so. How does the Nyquist frequency relate to 2 dimensions? I already am sampling such that I take a measurement about 1 per 20 micron, so this is way over the 2x that the nyquist frequency represents.

I am already performing deconvolution of the image in the spacial freq. domain after FFTs and IFFTs, I think (but really have no idea) that there's a way to do some anti-aliasing corrections in the sf domain as well when i perform the deconvolution, anyone have any insight into the anti-aliasing corrections of an image?


Thanks,

Jonathan
0 Kudos
Message 5 of 6
(3,218 Views)
Two points:
  • You are confusing the anti-aliasing needed for good data acquisition with the anti-aliasing methods used by images. From your description, you probably don't need to be concerned about either. If you data collection apparatus is fast enough, you won't get data acquisition anti-aliasing artifacts. Since your features are well within the resolution of your apparatus, this should not be a problem. Image anti-aliasing is something you do to make diagonal/curved lines on an image look smoother. As a simple example, take a two pixel wide black line running diagonally down a white background on an image. You can either turn a pixel black when more than 50% of it would be covered by the line (no aliasing) or set pixels to a gray value based on how much of the pixel is covered by the line (anti-aliasing). This technique has no real relevance to your work.

  • You are saving your data as a JPEG. While this saves you lots of space, it also distorts your data and throws away a lot of information. JPEG is a lossy storage format, designed and suitable only for photo-like images. JPEG compression seriously attenuates high frequency information, thus produces “ringing” effects around any sharp boundary, making it unsuitable for line drawings and any sort of data collection/analysis. The PNG format is a much better choice (as is TIFF, but TIFF is less well supported). PNG uses lossless compression, so is ideal for experimental data. PNG compression is extremely good for line drawings, as well. It will not compress as much as JPEG, but you won't lose any information either.
Hopefully this hasn't been too terse. If you need more info, repost and I'll try again.
0 Kudos
Message 6 of 6
(3,187 Views)